Will AI replace your job? Which careers are most at risk, and what skills matter now

Recently, Canva, the Australian design software company, said this week that more than a fifth of its employees would lose their jobs, change roles or relocate as it reorganises around artificial intelligence. In the same week NICE, a maker of customer-service software, announced about 500 job cuts. Across the corporate world, executives reporting quarterly results have been delivering a version of the same message to investors: AI will help them run leaner.

The layoffs are the visible part of a broader change. Inside jobs that still exist, AI has been reportedly taking over the work piece by piece: the summarising, sorting and drafting that once filled a working day.

So, which careers are most exposed, and which skills will hold their value?

Experts weigh in.

Most of us think about AI and jobs in all-or-nothing terms: Either a robot takes your seat or it doesn't. Fahed Bizzari, managing partner at Bellamy Alden, says that's the wrong place to start. "AI takes on the tasks inside a job one at a time, so the first step is to understand how exposed a given task is."

His test comes down to two questions. "First, can the task be taught to a capable newcomer? If so, AI can learn it too. Second, does it recur? The more often it recurs, the more incentive there is to invest in teaching it."

Skill alone, is not a shield. "A lawyer combing through their opponent's documents for contradictions, gaps in the evidence and anything else that weakens the other side's case is doing highly skilled work," Bizzari says. "But they're looking for the same kinds of weakness in every case, which means it can be taught to an AI. That's also why, as Anthropic's research found earlier this year, the people most exposed to AI tend to be better educated and better paid."

So, which jobs are most exposed to AI? Clerical work, bookkeeping, translation, first-line customer support and routine writing are among the roles, Bizzari identifies.

Dr Soumaya Askri, associate professor of marketing at Canadian University Dubai, points to similar areas, including data entry, administrative support, routine customer service, bookkeeping as well, and formulaic content production.

What makes these jobs vulnerable is the nature of the work itself, explains Musata Matei, partnerships lead at Women in AI and VP of ventures, partnerships and engagement at the AI innovation hub of a leading MENA bank.

"What these roles have in common is that a significant proportion of their work involves predictable, repetitive and information-heavy tasks: reading, sorting, summarising, drafting, processing documents or checking information against established rules."

The impact could be widespread. "The International Labour Organization (ILO) estimates that approximately one in four jobs globally is exposed to generative AI, with clerical occupations among the most exposed," Matei says.

Women could feel this shift more acutely, given their representation in many clerical roles. The ILO estimates that "9.6% of women's jobs in higher-income countries fall into the highest exposure category, compared with 3.5% of men's."

However, Matei cautions against viewing this as an inevitable wave of job losses. "My concern is not that AI will inevitably displace women, but that women could be disproportionately affected if we fail to prepare them for the jobs and opportunities being created."

There is also an important distinction between jobs being exposed to AI and jobs disappearing altogether. As Askri points out, "These are categories at risk of significant task automation or changes in staffing needs, not a prediction that every job in them will disappear."

How much can AI really do today? More than most people assume, says Bizzari. "With the right instruction, general-purpose AI can do almost any type of knowledge work, and anyone suggesting otherwise simply has not yet learned how to harness it."

One of his clients now uses AI to write 90 per cent of its proposals, cutting a three-week process to three days while also increasing its win rate, as he explains.

Inside banking, Matei is watching the same shift, with customer onboarding, KYC document processing, transaction monitoring and customer service, which are all increasingly benefiting from AI-driven automation and intelligent agents.

However, she draws a line every worker should understand. "AI is very good at producing a first answer, but that does not automatically make it the right answer," she says. "It can dramatically accelerate the first 70–80% of many tasks. The remaining part often requires human judgement, contextual understanding, accuracy and accountability."

Askri sees that gap clearly in the region. "In the UAE, for example, understanding local consumer behaviour and multilingual cultural context is not reducible to generating a campaign draft."

Bizzari offers a simple way to see where people fit. "In human-first work, people do the work and AI helps them."That covers anything that needs a body in the room or a name on the decision: "Work that needs hands, physical presence, relationships or someone to answer for the result will stay human-first."

The flip side is AI-first work, where "AI does the work under the instruction, guidance, supervision and engineering of people." His own firm has made that switch in corporate training. "While our AI tutor now teaches on our behalf, we design the courses, set the standards, review the results and keep improving how the AI teaches." People haven't left the building. Their job has moved from doing the work to running the thing that does it.

Are companies replacing people with AI, or expecting existing employees to do more? The answer is both.

Bizzari points to two different approaches. Walmart expects its workforce to remain the same size even as the business grows, meaning employees will take on more work. Other companies are cutting jobs. "This is why Salesforce reduced its customer support from about 9,000 people to about 5,000."

The impact on staffing can be significant. "If it stays human-first, a team that becomes five times more productive either needs a fifth of the people or five times as much work."

At the same time, companies are creating new roles around AI. "For example, we're now hiring marketing engineers instead of marketing managers."

However, not every use of AI means replacing employees. Askri points to research showing how the technology can help people work more efficiently. "In a published field study of customer-support agents, generative AI assistance increased productivity by about 15% on average, demonstrating augmentation rather than automatic replacement."

The bigger picture is churn, not collapse. Citing the World Economic Forum's Future of Jobs Report 2025, Matei notes that while around 40 per cent of employers expect to cut staff where AI can automate tasks, the report also projects that 170 million new jobs could be created globally by 2030, against 92 million displaced, she says.

She calls the time AI frees up a 'productivity dividend', and says what companies do with it is the real test. "The question is whether organisations reinvest that dividend in their people, giving employees the opportunity to develop new skills and take on higher-value responsibilities, or simply use it to reduce costs."

As she says, "You cannot introduce AI into an organisation, expect employees to adapt overnight and then call it transformation."

The impact of AI could be felt most strongly by people just starting their careers. "Junior roles are particularly vulnerable because they're made up almost entirely of highly exposed tasks," says Bizzari.

There are already signs of a shift. In September 2025, PwC UK reduced its graduate and school-leaver intake from 1,500 to 1,300, citing AI among other pressures. Researchers at Stanford's Digital Economy Lab have also found that employment among 22- to 25-year-olds in the occupations most exposed to AI has fallen behind that of their peers in less-exposed fields.

For Bizzari, the concern goes beyond fewer entry-level jobs. "As a father of five, I find that very sad," he says. "Junior roles have traditionally been the first rung on career ladders where grunt work helped newcomers learn the profession."

There is also a longer-term question for employers: if fewer graduates are hired today, where will experienced professionals come from tomorrow? Matei raises this concern: "If organisations significantly reduce junior hiring because AI can perform many of these tasks, where will the next generation of experienced professionals come from?"

That also puts pressure on universities to rethink how they prepare students for work. Askri says: "Universities should assess the quality of students’ thinking, not merely whether they can produce a polished first draft."

For graduates, Matei suggests focusing on skills that go beyond simply using AI. "Knowing how to use AI is increasingly becoming a baseline skill. Knowing how to question it, validate it, improve it and apply it to solve an actual business problem is where the real value lies."

Flip the two-question test around and you find the safer ground. "The harder a task is to teach and the less often it comes round, the more resilient it is," says Bizzari. He sees three groups holding firm: judgement in very different situations, from leaders to negotiators and crisis managers; physical work in changing surroundings, such as nurses, electricians and on-site maintenance technicians; and roles built on relationships and trust, like counsellors, social workers and salespeople.

AI is also creating its own demand, from building and fine-tuning AI systems to constructing the data centres they run on. Matei adds AI governance, cybersecurity, risk and regulation to the list of growth areas.

The skills that protect these roles are often waved away as 'soft'. Matei has a different perspective. "I would argue that, in the AI economy, they are becoming some of the hardest and most valuable skills to find."

Still, nobody gets a free pass. As Askri puts it: "No occupation is entirely immune to change."

On the first move, the experts are unanimous. "The most important step everyone must take, without further delay, is to develop practical AI competence," says Bizzari. Matei agrees that you don't need to become a software engineer, but you do need to know how the tools work and where they fall short. The urgency persists: The World Economic Forum estimates that 39% of workers' existing skill sets will be transformed or become outdated by 2030.

From there, their advice adds up to a simple plan:

For anyone still nervous of the technology, Bizzari has some reassurance. "For all the technophobes out there, AI is infinitely simpler to work with than traditional software."

The bigger danger isn't the machine itself, Matei argues. It's who gets left without the means to use it. "The real risk is not simply that AI will replace certain jobs. It is that people who are not given access to the skills and opportunities created by AI will be left behind."